cache invalidation: strategies for keeping data fresh. dont you hate stale data, right?

hey everyone, i've been thinking a lot about cache invalidation lately. weve got a pretty standard setup with redis for caching api responses, but keeping the data fresh without overwhelming the backend is always a challenge. were currently using a time-based expiration, but it leads to stale data. has any of you had success with other invalidation strategies like event-driven invalidation or explicit invalidation via api calls? what are the trade-offs youve observed in terms of complexity, performance, and data freshness?